Exploratory Data Analysis Part 1
Ankita Guha, Kara Marsh
April 22, 2018
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(plyr)## -------------------------------------------------------------------------## You have loaded plyr after dplyr - this is likely to cause problems.
## If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
## library(plyr); library(dplyr)## -------------------------------------------------------------------------##
## Attaching package: 'plyr'## The following objects are masked from 'package:dplyr':
##
## arrange, count, desc, failwith, id, mutate, rename, summarise,
## summarizelibrary(ggplot2)
library(ggmap)
library(mapdata)## Loading required package: maps##
## Attaching package: 'maps'## The following object is masked from 'package:plyr':
##
## ozonelibrary(maps)
library(stringr)
library(viridis)## Loading required package: viridisLitelibrary(maptools)## Loading required package: sp## Checking rgeos availability: FALSE
## Note: when rgeos is not available, polygon geometry computations in maptools depend on gpclib,
## which has a restricted licence. It is disabled by default;
## to enable gpclib, type gpclibPermit()library(gpclib) ## General Polygon Clipper Library for R (version 1.5-5)
## Type 'class ? gpc.poly' for helplibrary(sp)
gpclibPermit()## Warning in gpclibPermit(): support for gpclib will be withdrawn from
## maptools at the next major release## [1] TRUEeda <- read.table("data/cleaned_AQ.csv", header = TRUE, sep = ",")Proposition 1 We would like to figure out pollutants distribution across various regions of US based on years.
Data Preparation To meet the purpose of Proposition 1, let’s try to scoop out some of the chunk of datas that we might need to focus on. And then let’s try to free some of the memory for the data frame which we aren’t using.
eda1 <- subset(eda, select = c(latitude,
longitude,
parameter_name,
method_name,
year,
arithmetic_mean,
arithmetic_standard_dev,
state_name))
rm(eda) #filter(eda1, !is.na(method_name) | method_name != "")Okay, so let’s check out the datas across all the 30 years seperately. For the purpose of the Project’s scope limitation we would like to see the data for the year of 1987 and 2017, which indicates the beggining and the end of the air pollutants data available respectively.
# filter(eda1,year == "1987")
filter(eda1,year == "1987") %>% head()## latitude longitude parameter_name
## 1 36.56689 -118.77732 Suspended particulate (TSP)
## 2 40.11222 -75.30917 Carbon monoxide
## 3 39.73643 -86.21332 Sulfur dioxide
## 4 41.61662 -87.14696 PM10 Total 0-10um STP
## 5 32.79119 -116.94209 Nitrogen dioxide (NO2)
## 6 39.31417 -76.61333 Ozone
## method_name year arithmetic_mean
## 1 MEMBRANE-SAMPLER - GRAVIMETRIC 1987 25.518519
## 2 INSTRUMENTAL - NONDISPERSIVE INFRARED 1987 0.670625
## 3 Missing Data From Kaggle 1987 11.685389
## 4 HI-VOL-SA321A - GRAVIMETRIC 1987 38.757576
## 5 INSTRUMENTAL - CHEMILUMINESCENCE 1987 53.455056
## 6 Missing Data From Kaggle 1987 0.035872
## arithmetic_standard_dev state_name
## 1 21.079283 California
## 2 0.591866 Pennsylvania
## 3 13.279600 Indiana
## 4 24.645018 Indiana
## 5 26.144252 California
## 6 0.018500 Maryland# filter(eda1,year == "1988")
# filter(eda1,year == "1989")
# filter(eda1,year == "1990")
# filter(eda1,year == "1991")
# filter(eda1,year == "1992")
# filter(eda1,year == "1993")
# filter(eda1,year == "1994")
# filter(eda1,year == "1995")
# filter(eda1,year == "1996")
# filter(eda1,year == "1997")
# filter(eda1,year == "1998")
# filter(eda1,year == "1999")
# filter(eda1,year == "2000")
# filter(eda1,year == "2001")
# filter(eda1,year == "2002")
# filter(eda1,year == "2003")
# filter(eda1,year == "2004")
# filter(eda1,year == "2005")
# filter(eda1,year == "2006")
# filter(eda1,year == "2007")
# filter(eda1,year == "2008")
# filter(eda1,year == "2009")
# filter(eda1,year == "2010")
# filter(eda1,year == "2011")
# filter(eda1,year == "2012")
# filter(eda1,year == "2013")
# filter(eda1,year == "2014")
# filter(eda1,year == "2015")
# filter(eda1,year == "2016")
# filter(eda1,year == "2017")
filter(eda1,year == "2017") %>% head() ## latitude longitude parameter_name
## 1 41.24749 -95.97314 PM2.5 - Local Conditions
## 2 39.56333 -76.20389 Wind Speed - Resultant
## 3 41.84104 -71.36097 Relative Humidity
## 4 41.59851 -87.34299 PM2.5 - Local Conditions
## 5 30.11030 -84.99030 Ozone
## 6 40.96911 -95.04495 Ozone
## method_name year
## 1 Met One BAM-1020 Mass Monitor w/VSCC - Beta Attenuation 2017
## 2 INSTRUMENTAL - VECTOR SUMMATION LEVEL 4 2017
## 3 INSTRUMENTAL - HYGROSCOPIC PLASTIC FILM 2017
## 4 R & P Model 2025 PM-2.5 Sequential Air Sampler w/VSCC - Gravimetric 2017
## 5 INSTRUMENTAL - ULTRA VIOLET 2017
## 6 Missing Data From Kaggle 2017
## arithmetic_mean arithmetic_standard_dev state_name
## 1 10.721395 4.680987 Nebraska
## 2 5.921050 3.231377 Maryland
## 3 71.152758 20.909690 Rhode Island
## 4 9.786667 5.223144 Indiana
## 5 0.041389 0.007580 Florida
## 6 0.037717 0.008269 Iowa# Filtering highest and lowest mean of pollutants data for the year 1987
eda1987 <- eda1 %>% filter(year == "1987") %>% arrange(desc(arithmetic_mean))
eda1987 %>% head()## latitude longitude parameter_name method_name
## 1 32.72727 -117.1545 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## 2 36.72523 -119.7513 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## 3 33.92899 -118.2107 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## 4 34.28040 -119.3146 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## 5 34.92442 -120.4018 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## 6 37.33967 -121.8886 Total hydrocarbons INSTRUMENTAL - FLAME IONIZATION
## year arithmetic_mean arithmetic_standard_dev state_name
## 1 1987 20435.528 8998.2441 California
## 2 1987 17617.544 15073.3133 California
## 3 1987 3443.692 1858.0015 California
## 4 1987 2560.900 1264.9616 California
## 5 1987 2237.532 666.1922 California
## 6 1987 2196.192 522.5010 Californiaeda1987 %>% arrange(arithmetic_mean) %>% head()## latitude longitude parameter_name
## 1 34.51165 -120.49989 Vertical Wind Speed
## 2 44.58591 -68.80476 Temperature Difference
## 3 41.71776 -86.90779 Temperature Difference
## 4 34.40277 -119.45845 Vertical Wind Speed
## 5 40.00889 -75.09778 Beryllium (TSP) STP
## 6 38.25257 -85.74802 Bromomethane
## method_name year
## 1 Instrumental - Electronic or machine average 1987
## 2 Instrumental - Electronic or machine average level 2 - level 1 1987
## 3 Instrumental - Electronic or machine average level 2 - level 1 1987
## 4 Instrumental - Electronic or machine average 1987
## 5 HI-VOL - EMISSION SPECTRA ICAP 1987
## 6 CANISTER SUBAMBIENT PRESSURE - MULTI DETECTOR GC 1987
## arithmetic_mean arithmetic_standard_dev state_name
## 1 -0.209663 0.766333 California
## 2 -0.179139 1.171996 Maine
## 3 -0.140885 0.906252 Indiana
## 4 -0.123121 0.209601 California
## 5 0.000000 0.000000 Pennsylvania
## 6 0.000000 0.000000 Kentucky# Filtering highest and lowest mean of pollutants data for the year 2017
eda2017 <- eda1 %>% filter(year == "2017") %>% arrange(desc(arithmetic_mean))
eda2017 %>% head()## latitude longitude parameter_name
## 1 41.79734 -71.41793 Particle Number Total COunt
## 2 41.80778 -71.41510 Particle Number Total COunt
## 3 39.75118 -104.98762 Methane
## 4 37.05822 -88.57251 Elapsed Sample Time
## 5 37.10258 -107.87022 Elapsed Sample Time
## 6 38.06503 -84.49761 Elapsed Sample Time
## method_name
## 1 T-API 651/TSI 3783 at 3.0 lpm and 0.6 um cutpoint - Water-Based Condensation particle counter
## 2 T-API 651/TSI 3783 at 3.0 lpm and 0.6 um cutpoint - Water-Based Condensation particle counter
## 3 6L SUBAMBIENT SS-CANISTER - PRECONCENTRATOR GC/FID
## 4 R & P Model 2025 PM-2.5 FEM Sequential AIr Sampler - Calculation
## 5 R & P Model 2000 PM-2.5 FEM Air Sampler - Calculation
## 6 R & P Model 2025 PM-2.5 FEM Sequential AIr Sampler - Calculation
## year arithmetic_mean arithmetic_standard_dev state_name
## 1 2017 26889.56 23587.4956 Rhode Island
## 2 2017 13367.08 7754.3927 Rhode Island
## 3 2017 3152.50 333.6382 Colorado
## 4 2017 1440.00 0.0000 Kentucky
## 5 2017 1440.00 0.0000 Colorado
## 6 2017 1440.00 0.0000 Kentuckyeda2017 %>% arrange(arithmetic_mean) %>% head()## latitude longitude parameter_name
## 1 39.33850 -120.17129 Average Ambient Temperature
## 2 43.66025 -70.26896 Average Ambient Temperature
## 3 42.68075 -73.75733 Ambient Min Temperature
## 4 44.39308 -73.85890 Average Ambient Temperature
## 5 42.49984 -96.39476 Ambient Min Temperature
## 6 40.81259 -96.68331 Ambient Min Temperature
## method_name year
## 1 R & P Model 2025 PM-2.5 FEM Sequential Air Sampler - Electronic 2017
## 2 R & P Model 2025 PM-2.5 FEM Sequential Air Sampler - Electronic 2017
## 3 R & P Model 2025 PM2.5 Sequent - Electronic 2017
## 4 R & P Model 2025 PM2.5 Sequent - Electronic 2017
## 5 R & P Model 2025 PM-2.5 FEM Sequential Air Sampler - Electronic 2017
## 6 R & P Model 2025 PM-2.5 FEM Sequential Air Sampler - Electronic 2017
## arithmetic_mean arithmetic_standard_dev state_name
## 1 -6.133333 2.912044 California
## 2 -5.950000 3.717078 Maine
## 3 -5.309091 7.255957 New York
## 4 -4.875000 6.170562 New York
## 5 -4.286207 6.582592 Iowa
## 6 -3.886667 5.763787 Nebraska# Creating Another Data Frame for comparing Mean Value of Pollutants Data Across the Year 1987 & 2017
eda2 <- eda1 %>% filter(year == "1987" | year == "2017") %>% arrange(desc(arithmetic_mean))sapply(eda1987, count)## latitude longitude parameter_name method_name year
## x Numeric,2556 Numeric,2558 factor,143 factor,114 1987
## freq Integer,2556 Integer,2558 Integer,143 Integer,114 6633
## arithmetic_mean arithmetic_standard_dev state_name
## x Numeric,6095 Numeric,6144 factor,54
## freq Integer,6095 Integer,6144 Integer,54So apparently looks like there are 143 pollutants listed and the number of tests done in that year of 1987 is 114. We need to next find out which are the top tests and the pollutants name which are mostly found in the year.
Histogram of Arithmetic Mean of Pollutants Distribution Across Entire US for 30 Years
ggplot(data = eda1) + geom_histogram(aes(x = arithmetic_mean), fill = "plum", color = "black") + labs(
x = "Pollutant Mean", # x axis title
y = "Actual Pollutants", # y axis title
title = "Distribution of pollutants across US for 30 years" # main title of figure
) + scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 35945 rows containing non-finite values (stat_bin).
Histogram of Arithmetic Mean of Pollutants Distribution Across Entire US for the Year of 1987
ggplot(data = eda1987) +
geom_histogram(aes(x = arithmetic_mean), fill = "steelblue", color = "black") +
labs(
x = "Pollutant Mean", # x axis title
y = "Actual Pollutants", # y axis title
title = "Distribution of pollutants across US for the year of 1987" # main title of figure
) + scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 317 rows containing non-finite values (stat_bin).
Histogram of Arithmetic Mean of Pollutants Distribution Across Entire US for the Year of 2017
ggplot(data = eda2017) +
geom_histogram(aes(x = arithmetic_mean), fill = "salmon", color = "black") +
labs(
x = "Pollutant Mean", # x axis title
y = "Actual Pollutants", # y axis title
title = "Distribution of pollutants across US for the year of 2017" # main title of figure
) +
scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 401 rows containing non-finite values (stat_bin).
So apparently from the histogram distribution it looks like the distribution of the Pollutants comparatively decreased over the 30 years span. Let’s check some other paramters to validate this point.
Let’s check the variation of the mean value of the pollutants across the dataframes.
Density Plots
ggplot(eda1, aes(x=arithmetic_mean)) +
geom_density(fill = "plum", alpha = 0.90) + scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning: Removed 35945 rows containing non-finite values (stat_density).
ggplot(eda1987, aes(x=arithmetic_mean)) +
geom_density(fill = "steelblue", alpha = 0.90) + scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning: Removed 317 rows containing non-finite values (stat_density).
ggplot(eda2017, aes(x=arithmetic_mean)) +
geom_density(fill = "salmon", alpha = 0.90) + scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning: Removed 401 rows containing non-finite values (stat_density).
Looks like the distribution of the pollutants got an almost similar patterns in the entire datasets to that of the year 2017. Howver for the year of 1987 the density of distribution of the pollutants looks small. The only difference in the pollution density distribution is that in the year of 1987 and 2017 where at one point some of the pollutants density seemed to be at higher than that of the 2017.
Kernel Density Overlaid on Histogram
# Spanned Over 30 Years
ggplot(eda1, aes(x=arithmetic_mean, y=..density..)) +
geom_histogram(fill="plum", color="grey60", size=.2) +
geom_density() +
scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 35945 rows containing non-finite values (stat_bin).## Warning: Removed 35945 rows containing non-finite values (stat_density).
# For Year of 1987
ggplot(eda1, aes(x=arithmetic_mean, y=..density..)) +
geom_histogram(fill="steelblue", color="grey60", size=.2) +
geom_density() +
scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 35945 rows containing non-finite values (stat_bin).## Warning: Removed 35945 rows containing non-finite values (stat_density).
# For Year of 2017
ggplot(eda1, aes(x=arithmetic_mean, y=..density..)) +
geom_histogram(fill="salmon", color="grey60", size=.2) +
geom_density() +
scale_x_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 35945 rows containing non-finite values (stat_bin).## Warning: Removed 35945 rows containing non-finite values (stat_density).
Violin Plot
ggplot(eda1, aes(x="Pollutants", y=arithmetic_mean)) +
geom_violin(fill = "plum", alpha = 0.8) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 35945 rows containing non-finite values (stat_ydensity).
ggplot(eda1987, aes(x="Pollutants", y=arithmetic_mean)) +
geom_violin(fill = "steelblue", alpha = 0.8) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 317 rows containing non-finite values (stat_ydensity).
ggplot(eda2017, aes(x="Pollutants", y=arithmetic_mean)) +
geom_violin(fill = "salmon", alpha = 0.8) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 401 rows containing non-finite values (stat_ydensity).
Boxplots
ggplot(eda1, aes(y=arithmetic_mean, x=1)) +
geom_boxplot(fill = "plum", alpha = 0.2) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 35945 rows containing non-finite values (stat_boxplot).
ggplot(eda1987, aes(y=arithmetic_mean, x=1)) +
geom_boxplot(fill = "steelblue", alpha = 0.2) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 317 rows containing non-finite values (stat_boxplot).
ggplot(eda2017, aes(y=arithmetic_mean, x=1)) +
geom_boxplot(fill = "salmon", alpha = 0.2) +
scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 401 rows containing non-finite values (stat_boxplot).
As seen from both the Boxplot it can be said that: For Entire Dataset: The Median of the entire Pollutant mean is located almost exactly in the middle of the Box indicating a uniform distribution of Pollutants across Entire US as collected over the span of 30 years.
For 1987 US Data: The Median of the Pollutants Mean seems to be little more closer to the Upper end of the Spectrum (i.e towards the top whisker) indicating a higher amount of Pollutants distribution across the entire US during the year of 1987.
For 2017 US Data: The Median of the Pollutants Mean seems to be way more closer to the Upper end of the Spectrum (i.e towards the top whisker) indicating a higher amount of Pollutants distribution across the entire US during the year of 2017. However the entire Box size is smaller as compared to that of both the entire dataset as well as for the dataset of 1987, which could be probbaly due to less captured data value in the year of 2017 as compared to 1987.
Let’s try to get a visual comparison of the pollutants data for both the year of 1987 and 2017
ggplot(eda2, aes(x=factor(year), y=arithmetic_mean, color=year)) +
geom_boxplot() +
labs(
x = "Year", # x axis title
y = "Pollutant Mean", # y axis title
title = "Pollutant Distribution Across Year 1987 & 2017" # main title of figure
) + scale_y_log10()## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous y-axis## Warning: Removed 718 rows containing non-finite values (stat_boxplot).
And we can say conclusively that Median of the Pollutant Mean is more closer to the left whisker or the Top Spectrum of the Box for both the year of 1987 and 2017. This indicates that the Median of the Pollutant Mean are at a higher side in the year of 2017 as compared to the year of 1987. However the entire difference in the size of both the Boxplots as seen in both the year of 1987 and 2017 seems to be different due to the distribution variation in the Mean of the Pollutants data so collected.
Let’s try to check which regions in US we have these pollutants data for.
Google Satellite Map
map1 <- get_map(location = c(lon = mean(eda1$longitude),
lat = mean(eda1$latitude)),
zoom = 4,
maptype = "satellite",
scale = 2)## Map from URL : http://maps.googleapis.com/maps/api/staticmap?center=37.962973,-94.686641&zoom=4&size=640x640&scale=2&maptype=satellite&language=en-EN&sensor=false#knitr::include_graphics('Documents/Projects/AirQuality/staticmap.png')
ggmap(map1) +
geom_point(data = eda1,
aes(x = longitude,
y = latitude,
fill = "red",
alpha = 0.8),
size = 5,
shape = 21) +
guides(fill=FALSE, alpha=FALSE, size=FALSE)## Warning: Removed 16732 rows containing missing values (geom_point).
Interestingly for the entire data sets looks like the pollutants data are collected from all over the US.
Let’s try to check the regions covered in the year of 1987.
map1987 <- get_map(location = c(lon = mean(eda1987$longitude),
lat = mean(eda1987$latitude)),
zoom = 4,
maptype = "satellite",
scale = 2)## Map from URL : http://maps.googleapis.com/maps/api/staticmap?center=38.676913,-92.553185&zoom=4&size=640x640&scale=2&maptype=satellite&language=en-EN&sensor=falseggmap(map1987) +
geom_point(data = eda1987, aes(x = longitude,
y = latitude,
fill = "red",
alpha = 0.8),
size = 5,
shape = 21) +
guides(fill=FALSE, alpha=FALSE, size=FALSE)## Warning: Removed 574 rows containing missing values (geom_point).
Let’s try to see the same for the year of 2017.
map2017 <- get_map(location = c(lon = mean(eda2017$longitude),
lat = mean(eda2017$latitude)),
zoom = 4,
maptype = "satellite",
scale = 2)## Map from URL : http://maps.googleapis.com/maps/api/staticmap?center=38.398833,-96.206211&zoom=4&size=640x640&scale=2&maptype=satellite&language=en-EN&sensor=falseggmap(map2017) +
geom_point(data = eda2017, aes(x = longitude,
y = latitude,
fill = "red",
alpha = 0.8),
size = 5,
shape = 21) +
guides(fill=FALSE, alpha=FALSE, size=FALSE)## Warning: Removed 96 rows containing missing values (geom_point).
Let’s try to visualize the States with the highest mean pollutants distribution.
ggplot(data = eda1) +
geom_histogram(aes(x = arithmetic_mean,
binwidth = 30000,
fill = state_name)) +
scale_x_log10() ## Warning: Ignoring unknown aesthetics: binwidth## Warning in self$trans$transform(x): NaNs produced## Warning: Transformation introduced infinite values in continuous x-axis## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 35945 rows containing non-finite values (stat_bin).
Okay, from the Histogram distribution, it looks like, some of the States such as Alabama, Alaska, Arizona have a higher level concentration of pollutants that are captured across the datas in these span of 30 years.
Let’s reuse some of the dataframes for our next level of analysis
write.csv(eda1, file = "data/eda1.csv")
write.csv(eda1987, file = "data/eda1987.csv")
write.csv(eda2017, file = "data/eda2017.csv")
write.csv(eda2, file = "data/eda2.csv") #Entire US with 2017 & 1987Let’s try to free some of the used spaces for better use with our memory and for being able to fit other EDA’s conviniently.
rm(map1, map1987, map2017, eda1, eda1987, eda2, eda2017)Conclusion
It is interesting to note that the Pollutants data so collected are not in the same units of measurement, so Normalization would have produced much optimized exploratory data analysis. But that would require much more in-depth subject matter expertise & time so that could form a future study itself.
In this study we viewed the pollutants distribution value across US based on the year of 1987 & 2017 as well as for the Pollutants distribution across the Entire US for a span of 30 years.
Also for ease of analysis we have only included the Arithmetic Mean of the Pollutants and have assumed that those are normalized across all the Pollutants so captured in our analysis. Because for obvious reasons different Pollutants or Air Constituent would have different threshold level to be considered as Pollutant.
For the Map to be more explicit in nature, one can increase or decrease the zoom argument to acheieve the desired level of Map visual. We have decided our zoom to be at a desired level, for the purpose of providing a suitable aerial view of all the 4 States that we are looking into.
Another interesting fact point that we learned while using ggmap() is that after a certain point of time, query used to fetch map data might not run, if a certain quota of fetching map API data from google is met. We came across an error something like: geocode failed with status OVER_QUERY_LIMIT, location = “michigan”, that means that we have run our code many times and hence the IP address has met it’s limit to use and fetch API data from Google. **Source: https://stackoverflow.com/questions/tagged/google-geocoding-api?page=4&sort=unanswered**